§ CAPABILITY

Project Management Office Setup

The smallest amount of governance that makes multi-team delivery legible — intake, prioritisation, a shared definition of done, and one view of what is actually in flight.

Who
Led by the founder, hands-on for the duration.
Request a call How the work runs

What you're seeing

Nobody can say what is currently in flight without asking around.
Usually means There is no portfolio view, so the answer is assembled differently each time it is requested. Work running that leadership does not know about is the usual first finding.
Priorities are set by whoever asks most persistently.
Usually means There is no intake path, so the queue is ordered by escalation rather than by value. It also means the people who are politest about it get the least.
Two teams both report a feature as done and it is not live.
Usually means Done means different things to different teams. Once that is true the portfolio view stops being believed, and shortly afterwards stops being maintained.
A stage gate exists that everybody routes around.
Usually means Governance was imported rather than designed, and it costs more than the risk it manages. A gate nobody respects is worse than none, because it teaches the organisation that process is theatre.

The smallest governance that works

Most PMO material is written for organisations with hundreds of engineers.

Imported wholesale by an organisation with a fraction of that, it produces reporting nobody reads and stage gates nobody respects — and it does lasting damage, because the company concludes that governance is theatre and stops trying. The correct response is not more discipline. It is less governance, designed rather than adopted.

What is actually needed at this size is small: a way for work to enter, someone who decides, a shared definition of done, and one current picture of what is running. Four things. Everything beyond them has to earn its place individually.

The test

Every artefact has to change a decision.

A register that is diligently updated and never consulted is a cost with no return, and it is remarkably common because updating it feels like diligence. Applying the test honestly means the most frequent output of this work is deletion: the weekly report that nobody acts on, the field in the tracker that four people fill in and none read, the approval step that has never once resulted in a no.

What survives is small and used, which is the only version that outlives the engagement.

One definition of done

The portfolio view stops being believed the moment two teams report complete against different standards.

One team means merged. Another means deployed behind a flag. A third means live, observable and documented. All three are defensible internally and the aggregate is meaningless, which is why portfolio dashboards decay — not from neglect, but because people stopped trusting them and quietly went back to asking.

Agreeing one definition, including deployed and observable, is the cheapest single thing that makes a cross-team view worth maintaining.

Where it sits

This capability sits inside Delivery Implementation, where multi-team governance is one of the tracks being installed ahead of a funding event, and under Engineering Velocity where the constraint turns out to be coordination across teams rather than delivery inside them.

It sits between two neighbours. Delivery Management owns an individual commitment; this is the layer that makes the set of them visible together. And Board Delivery Reporting is where the portfolio view is translated for people who are funding it rather than running it. Process Rebuild is what happens when the picture shows the problem is inside the teams after all.

How the work runs

  1. Inventory what is in flight

    Every initiative with an owner, a stage and a claimed date. On a first pass this usually surfaces work nobody at the top of the company knew was running.

  2. Design intake and prioritisation

    How work enters, who decides, and against what. Without an intake path, prioritisation happens by whoever asks most persistently.

  3. Agree a shared definition of done

    One that includes deployed, observable and documented. Teams reporting complete against different definitions is why portfolio views stop being believed.

  4. Build the portfolio view and hand it over

    A single current picture generated from the delivery tools, owned by someone inside the company rather than by us.

What arrives

  • A portfolio inventory with owner, stage and date per initiative
  • A written intake and prioritisation path with named decision rights
  • A shared definition of done applied across teams
  • One portfolio view generated from existing tooling, with an internal owner

What it costs your team

Two workshops plus about half a day a month from leadership once it is running.

Next step

Tell us where this is breaking and we'll say what it takes to fix.

Request a call

How we decide

  • Every artefact has to change a decision, or it is cut

    Costs It removes reports that some stakeholders have grown attached to.

    This is the test that separates governance from bureaucracy, and it is the only one that holds. A register that is updated and never consulted has a real cost and no return. Applying the test honestly usually means the most common output of this work is deletion rather than addition.

  • The size follows the company, not the template

    Costs It produces something an enterprise PMO practitioner would consider incomplete.

    Most PMO material is written for organisations with hundreds of engineers and gets imported wholesale by organisations with far fewer. The result is reporting nobody reads and stage gates nobody respects, which discredits the idea of governance entirely for that company — usually for years.

  • It is handed to an internal owner as part of the scope

    Costs It ends a piece of work that could reasonably continue.

    A governance function that depends on an outside party has failed at the thing it exists to do. The intake path, the prioritisation forum and the portfolio view all have to survive us leaving, which means someone inside the company runs them from early on rather than inheriting them at the end.

Frequently Asked Questions

A project management office — the function that decides how work enters an organisation, how it is prioritised against other work, and how its status is reported. In a large company that is a department with staff. In a scaling one it is a practice: the smallest set of agreements that makes multi-team delivery legible to the people funding it.
They become bureaucracy when the governance outweighs the delivery it governs, which is the reliable outcome of importing an enterprise template into a company that does not need one. The test applied here is whether each artefact changes a decision. Anything that only produces a report gets cut, and that is usually most of it.
When more than a few teams share dependencies and nobody can answer what is in flight without asking around. Below that threshold a shared document does the job, and we will say so rather than build something that justifies an engagement.
No, and the distinction is worth being clear about. Enterprise PMO work is a different service with a different buyer, a heavier framework and a much larger governance footprint. This is sized for a company that has outgrown one team and has not become a corporation, and it is deliberately light.
Someone inside the company, and that person is identified at the start rather than at the end. Handover is part of the scope for a structural reason: a governance practice that only functions while an outside party is present has not been established, it has been performed.

Sources

Page reviewed